Here is a super simple statistics question whose answer I didn’t know until just now: How do you test whether one integer is different from another?

Like, I do two things and count 3 the first time and 5 the second time. Did I do the same thing both times? I dunno, maybe I did.

But if I counted 20 the first time and 50 the second time, I’m starting to feel like the things are different. And if I counted 100000 one time and 10 the other, the things seem very different indeed.

The right way to formalize this is apparently to use the Poisson distribution family. You treat each count as a measurement of the rate at which one process does something. And then you can do statistical tests to see whether the two different counts seem to come from processes with different rates.

You can see a calculator for this here.

I’m unsettled by how often I accidentally think of really basic-seeming statistics questions which I can’t answer :pensive: